Sentiment analysis is an important business intelligence tool that helps companies improve their products and services by giving further insights to calls. Some benefits include:
- Helping managers identify areas of improvement and current bottlenecks in Customer Support
- Helping agents gauge the mood of their last customer call and adapt their tone for the next one
- Quickly identifying calls that need attention by filtering for ones with negative sentiment
- Analyzing at scale across all calls
- Avoiding personal biases when analyzing customer interactions
Who has access?
Users subscribed to the Aircall AI add-on will have access to sentiment analysis.
Analysis will be available to those users for calls longer than 1 minute which have transcripts in English (EN), French (FR), Spanish (ES), and/or German (DE).
The analysis will be available in the Conversation Center and on each recording/voicemail page.
The analysis will be available on future calls from the moment the feature is enabled/rolled out, but will not be available for calls that occurred prior.
Functionality
What is sentiment analysis?
Sentiment analysis detects a customer’s mood after a conversation. Aircall's sentiment analysis is based on language models that analyze call transcripts with emotion and sentiment models.
⚠️ Please note that since Aircall's sentiment analysis is based on transcripts, no audio from calls will be accounted for in the sentiment.
How can the users interact with this analysis
- View the general customer mood for each conversation
- View what specific parts of each transcript are considered positive, negative, or neutral
- Visualize the customer's mood through the timeline of a call
Legal and Compliance
For some users, depending on the vertical and geography they operate in, using this analysis may trigger compliance obligations. Should you wish to disable the AI Sentiment Analysis feature, please create a support ticket and the analysis will be disabled for your entire account.
What are some limitations of the analysis?
- It is based on the text of transcripts, therefore the tone of voice is not evaluated based on audio, and nuances/precision might get lost because of that.
- Sentiment is purely about the customer's mood. We can identify customers with bad moods, but we will not identify higher level things such as: is this call good or bad for our company? For example, a customer with a bad mood about a competitor will have a negative sentiment, even though this may be good for your company.
- Aircall will not take into account the outcome of the call.